On the generalization capabilities of FSL methods through domain adaptation: a case study in endoscopic kidney stone image classification

TL;DR

This paper investigates the generalization of few-shot learning methods in medical image classification, demonstrating that meta-learning approaches outperform traditional deep learning in cross-domain kidney stone image classification tasks.

Contribution

It introduces a meta-learning based few-shot learning approach to improve domain generalization in medical image classification, validated on endoscopic kidney stone datasets.

Findings

Meta-learning methods achieve 74.38% and 88.52% accuracy in 5-way 5-shot and 5-way 20-shot tasks.

Traditional deep learning methods attain only 45% accuracy under the same conditions.

Meta-learning effectively handles domain shifts in medical imaging.

Abstract

Deep learning has shown great promise in diverse areas of computer vision, such as image classification, object detection and semantic segmentation, among many others. However, as it has been repeatedly demonstrated, deep learning methods trained on a dataset do not generalize well to datasets from other domains or even to similar datasets, due to data distribution shifts. In this work, we propose the use of a meta-learning based few-shot learning approach to alleviate these problems. In order to demonstrate its efficacy, we use two datasets of kidney stones samples acquired with different endoscopes and different acquisition conditions. The results show how such methods are indeed capable of handling domain-shifts by attaining an accuracy of 74.38% and 88.52% in the 5-way 5-shot and 5-way 20-shot settings respectively. Instead, in the same dataset, traditional Deep Learning (DL) methods attain only an accuracy of 45%.